The paper is devoted to the project Nr AM 2007/52 funded by Ministry of Defense of Republic of Latvia. The paper describes the main results and problems that were studied within the project. The main goal of the project was to develop a model of software system that controls autonomous robotic system. The most characterizing feature of autonomous systems is their mobility and ability to plan their actions in unknown and dynamic environments. Therefore most of the project efforts were related to navigation and motion planning problems. The navigation itself consists of several sub-problems including self-localization, mapping and high level planning. To address all of these problems the well known three-layer architecture ATLANTIS was adopted and slightly modified without loosing the advantages of heterogeneous control in order to fit the technical solution of the software developed within the project. The self-localization is addressed by applying laser scanner data processing algorithm proposed by Martinez. The laser scanner is the leading sensor that is used for robot navigation. Motion planning is addressed by applying RRT planner that is well suited for discreet motion planning. For high level task planning specialized algorithms were built that allow to investigate closed environment and patrol within the closed area by providing several control point on a metric map. The main tool for interacting with user is geographic information system which data is provided by the robotic system thereby addressing the mapping problem. The robotic system is implemented as 3D visual model. The further research efforts are related with development of operational prototype, development of appropriate mathematical model and control of continuous motion as well as further advancement of integrated control software for high level task planning.